The purpose of this study is to test the generalizability of a community-based intervention (Be Proud! Be Responsible!) (BPBR) designed to reach younger, inner-city adolescents (Jemmott, Jemmott, Fong, 1998) by taking the program into health classes of 5 urban and suburban high schools, targeting older adolescents (9th/10th graders), and comparing results to alternative health promotion intervention offered in 5 other schools. The primary goal of the project is to determine if the culturally sensitive BPBR intervention will be as effective when conducted: (a) within a high school health curriculum, and (2) among white, as well as African-American suburban youth. The secondary goal is to extend the theoretical framework underlying the intervention to explore contextual factors (individual, classroom, facilitator and school) in order to better understand the conditions under which the intervention is most effective on changing behaviors (sexual activity, number of partners, condom use, sex initiation) and the mediators theoretically assumed to precede behavior change (e.g., knowledge, prevention/hedonistic beliefs, intentions, efficacy). We propose a group-randomized trial with a nested cohort design, whereby five pairs of schools that are similar in SES and racial composition of the student population will be randomized to the intervention or control group. In addition, within each school, classrooms will be randomized to one of the two facilitators: health education teacher or school nurse. Finally, 25% of the classes will be selected to be taught in same gender classrooms. Data will be collected at the individual (e.g., sociodemographics, behavioral outcomes, mediators), classroom (e.g., age/race/sex composition, level of teacher's control, program fidelity, facilitator (HE teacher versus nurse, comfort level with material), and school (urban versus suburban racial mix) level at pre-intervention and immediately following the intervention classes. Individual data will again be collected at 3 months, 6 months and 12 months following the intervention. A booster session will be held in randomly selected schools to determine the importance of such sessions to sustain the intervention effects. The data will be analyzed using an array of analytic techniques including ANOVAs, OLS, path modeling and finally, hierarchical modeling which will permit the examination of change over time within a nested design (individuals within classes within schools). The study will provide valuable information with regard to replication and generalizability of the intervention, but also a better understanding of the underlying mechanisms that predict what adolescents think and do.